Quasar Identification Using Multivariate Probability Density Estimated from Nonparametric Conditional Probabilities

Nonparametric estimation for a probability density function that describes multivariate data has typically been addressed by kernel (KDE). A novel estimator recently developed Farmer and Jacobs offers an alternative high-throughput automated approach to univariate nonparametric based on maximum entropy order statistics, improving accuracy over KDE. This article presents extension of the single variable case multiple variables. The is used recursively calculate product array one-dimensional conditional probabilities. In combination with interpolation methods, complete joint estimate generated Good speed performance in synthetic are demonstrated numerical study using known distributions range sample sizes from 100 106 two six Performance terms compared here tends perform better as number samples and/or variables increases. As example application, measurements analyzed five filters photometric Sloan Digital Sky Survey Data Release 17. form basis binary classifier distinguishes quasars galaxies stars up 94% accuracy.